Mathematical Intuition: Poincaré, Pólya, Dewey

Similar documents
Remarks on the philosophy of mathematics (1969) Paul Bernays

Class #14: October 13 Gödel s Platonism

On the epistemological status of mathematical objects in Plato s philosophical system

Rethinking Knowledge: The Heuristic View

McDougal Littell High School Math Program. correlated to. Oregon Mathematics Grade-Level Standards

Mathematics as we know it has been created and used by

Pictures, Proofs, and Mathematical Practice : Reply to James Robert Brown

ASPECTS OF PROOF IN MATHEMATICS RESEARCH

Lecture 9. A summary of scientific methods Realism and Anti-realism

CONTENTS A SYSTEM OF LOGIC

MISSOURI S FRAMEWORK FOR CURRICULAR DEVELOPMENT IN MATH TOPIC I: PROBLEM SOLVING

Semantic Foundations for Deductive Methods

Proof as a cluster concept in mathematical practice. Keith Weber Rutgers University

It Ain t What You Prove, It s the Way That You Prove It. a play by Chris Binge

Brief Remarks on Putnam and Realism in Mathematics * Charles Parsons. Hilary Putnam has through much of his philosophical life meditated on

Rational and Irrational Numbers 2

III Knowledge is true belief based on argument. Plato, Theaetetus, 201 c-d Is Justified True Belief Knowledge? Edmund Gettier

Laws are simple in nature. Laws are quantifiable. Formulated laws are valid at all times.

AKC Lecture 1 Plato, Penrose, Popper

Logic: Deductive and Inductive by Carveth Read M.A. CHAPTER VI CONDITIONS OF IMMEDIATE INFERENCE

Lecture 6. Realism and Anti-realism Kuhn s Philosophy of Science

Science, Rationality and the Human Mind. by Garry Jacobs

- We might, now, wonder whether the resulting concept of justification is sufficiently strong. According to BonJour, apparent rational insight is

9 Knowledge-Based Systems

"We Think That We Think Clearly, But That's Only Because We Don't Think Clearly": Brian Josephson on Mathematics, Mind, and the Human World*

COMPOSITIO MATHEMATICA

Grade 6 correlated to Illinois Learning Standards for Mathematics

Predicate logic. Miguel Palomino Dpto. Sistemas Informáticos y Computación (UCM) Madrid Spain

Here s a very dumbed down way to understand why Gödel is no threat at all to A.I..

Probability Foundations for Electrical Engineers Prof. Krishna Jagannathan Department of Electrical Engineering Indian Institute of Technology, Madras

PHILOSOPHY 4360/5360 METAPHYSICS. Methods that Metaphysicians Use

Conventionalism and the linguistic doctrine of logical truth

Why Rosenzweig-Style Midrashic Approach Makes Rational Sense: A Logical (Spinoza-like) Explanation of a Seemingly Non-logical Approach

"Can We Have a Word in Private?": Wittgenstein on the Impossibility of Private Languages

6.041SC Probabilistic Systems Analysis and Applied Probability, Fall 2013 Transcript Lecture 21

Kant Lecture 4 Review Synthetic a priori knowledge

The Development of Laws of Formal Logic of Aristotle

Logic: Deductive and Inductive by Carveth Read M.A. CHAPTER IX CHAPTER IX FORMAL CONDITIONS OF MEDIATE INFERENCE

BEGINNINGLESS PAST AND ENDLESS FUTURE: REPLY TO CRAIG. Wes Morriston. In a recent paper, I claimed that if a familiar line of argument against

It doesn t take long in reading the Critique before we are faced with interpretive challenges. Consider the very first sentence in the A edition:

Ayer s linguistic theory of the a priori

Bayesian Probability

LOCKE STUDIES Vol ISSN: X

Introductory Kant Seminar Lecture

Is There an External World? George Stuart Fullerton

Full-Blooded Platonism 1. (Forthcoming in An Historical Introduction to the Philosophy of Mathematics, Bloomsbury Press)

PHI2391: Logical Empiricism I 8.0

Cartesian Rationalism

REFUTING THE EXTERNAL WORLD SAMPLE CHAPTER GÖRAN BACKLUND

Perceiving Abstract Objects

Broad on Theological Arguments. I. The Ontological Argument

Kant and his Successors

Content Area Variations of Academic Language

The Unbearable Lightness of Theory of Knowledge:

Descartes and Foundationalism

This handout follows the handout on The nature of the sceptic s challenge. You should read that handout first.

World without Design: The Ontological Consequences of Natural- ism , by Michael C. Rea.

Ayer and Quine on the a priori

Number, Part I of II

On Infinite Size. Bruno Whittle

KRIPKE ON WITTGENSTEIN. Pippa Schwarzkopf

THE ROLE OF COHERENCE OF EVIDENCE IN THE NON- DYNAMIC MODEL OF CONFIRMATION TOMOJI SHOGENJI

Cartesian Rationalism

The Development of Knowledge and Claims of Truth in the Autobiography In Code. When preparing her project to enter the Esat Young Scientist

HUME'S THEORY. THE question which I am about to discuss is this. Under what circumstances

REASONING ABOUT REASONING* TYLER BURGE

EPISTEMOLOGY for DUMMIES

LTJ 27 2 [Start of recorded material] Interviewer: From the University of Leicester in the United Kingdom. This is Glenn Fulcher with the very first

Philosophy of Science. Ross Arnold, Summer 2014 Lakeside institute of Theology

WHAT IS HUME S FORK? Certainty does not exist in science.

6.041SC Probabilistic Systems Analysis and Applied Probability, Fall 2013 Transcript Lecture 3

On The Logical Status of Dialectic (*) -Historical Development of the Argument in Japan- Shigeo Nagai Naoki Takato

Experience and Foundationalism in Audi s The Architecture of Reason

1. Introduction Formal deductive logic Overview

Writing Module Three: Five Essential Parts of Argument Cain Project (2008)

Ayer on the criterion of verifiability

I Found You. Chapter 1. To Begin? Assumptions are peculiar things. Everybody has them, but very rarely does anyone want

The Kripkenstein Paradox and the Private World. In his paper, Wittgenstein on Rules and Private Languages, Kripke expands upon a conclusion

Putnam on Methods of Inquiry

UNIVALENT FOUNDATIONS

2.1 Review. 2.2 Inference and justifications

Curriculum Guide for Pre-Algebra

Illustrating Deduction. A Didactic Sequence for Secondary School

Religion and Science: The Emerging Relationship Part II

Artificial Intelligence Prof. Deepak Khemani Department of Computer Science and Engineering Indian Institute of Technology, Madras

KANT S EXPLANATION OF THE NECESSITY OF GEOMETRICAL TRUTHS. John Watling

Is Mathematics! Invented! OR! Discovered?!

Reid Against Skepticism

GREAT PHILOSOPHERS: Thomas Reid ( ) Peter West 25/09/18

Introduction Symbolic Logic

Does Deduction really rest on a more secure epistemological footing than Induction?

Meaning of the Paradox

15 Does God have a Nature?

From Transcendental Logic to Transcendental Deduction

Ch01. Knowledge. What does it mean to know something? and how can science help us know things? version 1.5

The Ontological Argument for the existence of God. Pedro M. Guimarães Ferreira S.J. PUC-Rio Boston College, July 13th. 2011

OSSA Conference Archive OSSA 3

K.V. LAURIKAINEN EXTENDING THE LIMITS OF SCIENCE

Gödel's incompleteness theorems

1.2. What is said: propositions

Transcription:

Mathematical Intuition: Poincaré, Pólya, Dewey Reuben Hersh Abstract Practical calculation of the limit of a sequence often violates the definition of convergence to a limit as taught in calculus. Together with examples from Euler, Pólya and Poincaré, this fact shows that in mathematics, as in science and in everyday life, we are often obligated to use knowledge that is derived, not rigorously or deductively, but simply by making the best use of available information plausible reasoning. The philosophy of mathematical practice fits into the general framework of warranted assertibility, the pragmatist view of the logic of inquiry developed by John Dewey. Keywords Intuition Induction Pragmatism Approximation Convergence Limits Knowledge 1 Introduction In Rio de Janeiro in May 2010, I spoke at a meeting of numerical analysts honoring the 80th anniversary of the famous paper by Courant, Friedrichs, and Lewy. In order to give a philosophical talk appropriate for hard-core computer-oriented mathematicians, I focused on a certain striking paradox that is situated right at the heart of analysis, both pure and applied. (That paradox was presented, with considerable mathematical elaboration, in Phil Davis s excellent article, The Paradox of the Irrelevant Beginning, cf.[5].) In order to make this paradox cut as sharply as possible, I performed a little dialogue, with help from Carlos Motta. With the help of Jody Azzouni, I used that dialogue again, to introduce this talk in Rome. To set the stage, recall the notion of a convergent sequence, which is at the heart of both pure analysis and applied mathematics. In every calculus course, the student learns that whether a sequence converges to a limit and what that limit is depend only on the end of the sequence that is, the part that is very far out in the tail, so to speak, or in the infinite part. Yet, in a specific instance when the limit is R. Hersh ( ) Department of Mathematics and Statistics, University of New Mexico, Albuquerque, USA e-mail: rhersh@gmail.com C.A. de Moura, C.S. Kubrusly (eds.), The Courant Friedrichs Lewy (CFL) Condition, DOI 10.1007/978-0-8176-8394-8_2, Springer Science+Business Media New York 2013 9

10 R. Hersh actually needed, usually all that is considered is the beginning of the sequence the first few terms the finite part, so to speak. (Even if the calculation is carried out to a hundred or a thousand iterations, this is still only the first few, compared to the remaining, neglected, infinite tail.) In this little drama of mine, the hero is a sincere, well-meaning student, who has not yet learned to accept life as it really is. A second character is the Successful Mathematician the Ideal Mathematician s son-in-law. His mathematics is ecumenical: a little pure, a little applied, and a little in-between. He has grants from federal agencies, a corporation here and there, and a private foundation or two. His conversation with the Stubborn Student is somewhat reminiscent of a famous conversation between his Dad, the Ideal Mathematician, and a philosophy grad student, who long ago asked, What is a mathematical proof, really? 2 A Dialogue The Successful Mathematician (SM) is accosted by the Stubborn Student (SS) from his Applied Analysis course. SS: Sir, do you mind if I ask a stupid question? SM: Of course not. There is no such thing as a stupid question. SS: Right. I remember, you said that. So here s my question. What is the real definition of convergence? Like, convergence of an infinite sequence, for instance? SM: Well, I m sure you already know the answer. The sequence converges to a limit, L, if it gets within a distance epsilon of L, and stays there, for any positive epsilon, no matter how small. SS: Sure, that s in the book, I know that. But then, what do people mean when they say, keep iterating till the iteration converges? How does that work? SM: Well, it s obvious, isn t it? If after a hundred terms your sequence stays at 3, correct to four decimal places, then the limit is 3. SS: Right. But how long is it supposed to stay there? For a hundred terms, for two hundred, for a hundred million terms? SM: Of course, you wouldn t go on for a hundred million. That really would be stupid. Why would you waste time and money like that? SS: Yes, I see what you mean. But what then? A hundred and ten? Two hundred? A thousand? SM: It all depends on how much you care. And how much it is costing, and how much time it is taking. SS: All right, that s what I would do. But when does it converge? SM: I told you. It converges if it gets within epsilon... SS: Never mind about that. I am supposed to go on computing until it converges, so how am I supposed to recognize that it has converged?

Mathematical Intuition: Poincaré, Pólya, Dewey 11 SM: When it gets within four decimal points of some particular number and stays there. SS: Stays there how long? Till when? SM: Whatever is reasonable. Use your judgment! It s just plain common sense, for Pete s sake! SS: But what if it keeps bouncing around within four decimal points and never gets any closer? You said any epsilon, no matter how small, not just 0.0001. Or if I keep on long enough, it might finally get bigger than 3, even bigger than 4, way, way out, past the thousandth term. SM: Maybe this, maybe that. We haven t got time for all these maybes. Somebody else is waiting to get on that machine. And your bill from the computing center is getting pretty big. SS: (mournfully) I guess you re not going to tell me the answer. SM: You just don t get it, do you? Why don t you go bother that Reuben Hersh over there, he looks like he has nothing better to do. SS: Excuse me, Professor Hersh. My name is RH: That s OK. I overheard your conversation with Professor Successful over there. Have a seat. SS: Thank you. So, you already know what my question is. RH: Yes, I do. SS: So, what is the answer? RH: He told you the truth. The real definition of convergence is exactly what he said, with the epsilon in it, the epsilon that is arbitrarily small but positive. SS: So then, what does it mean, go on until the sequence converges, then stop? RH: It s meaningless. It s not a precise mathematical statement. As a precise mathematical statement, it s meaningless. SS: So, if it s meaningless, what does it mean? RH: He told you what it means. Quit when you can see, when you can be pretty sure, what the limit must be. That s what it means. SS: But that has nothing to do with convergence! RH: Right. SS: Convergence only depends on the last part, the end, the infinite part of the sequence. It has nothing to do with the front part. You can change the first hundred million terms of the sequence, and that won t affect whether it converges, or what the limit is. RH: Right! Right! Right! You really are an A student. SS: I know... So it all just doesn t make any sense. You teach us some fancy definition of convergence, but when you want to compute a number, you just forget about it and say it converges when common sense, or whatever you call it, says something must be the answer. Even though it might not be the answer at all! RH: Excellent. I am impressed.

12 R. Hersh SS: Stop patronizing me. I m not a child. RH: Right. I will stop patronizing you because you are not a child. SS: You re still doing it. RH: It s a habit. I can t help it. SS: Time to break a bad habit. RH: OK. But seriously, you are absolutely right. I agree with every word you say. SS: Yes, and you also agree with every word Professor Successful says. RH: He was telling the truth, but he couldn t make you understand. SS: All right. You make me understand. RH: It s like theory and practice. Or the ideal and the actual. Or Heaven and Earth. SS: How is that? RH: The definition of convergence lives in a theoretical world. An ideal world. Where things can happen as long as we can clearly imagine them. As long as we can understand and agree on them. Like really being positive and arbitrarily small. No number we can write down is positive and arbitrarily small. It has to have some definite size if it is actually a number. But we can imagine it getting smaller and smaller and smaller while staying positive, and we can even express that idea in a formal sentence, so we accept it and work with it. It seems to convey what we want to mean by converging to a limit. But it s only an ideal, something we can imagine, not something we can ever really do. SS: So you re saying mathematics is all a big fairy tale, a fiction, it doesn t actually exist? RH: NO! I never said fairy tale or fiction. I said imaginary. Maybe I should have said consensual. Something we can all agree on and work with because we all understand it the same way. SS: That s cool. We all. All of you. Does that include me? RH: Sure. Stay in school a few more years. Learn some more. You ll get into the club. You ve got what it takes. SS: I m not so sure. I have trouble believing two opposite things at once. RH: Then how do you get along in daily life? How do you even get out of bed in the morning? SS: What are you talking about? RH: How do you know someone hasn t left a bear trap by your bedside that will chop off your foot as soon as you step down? SS: That s ridiculous. RH: It is. But how do you know it is? SS: Never mind how I know. I just know it s ridiculous. And so do you. RH: Exactly. We know stuff, but we don t always know how we know it. Still, SS: we do know it. So you re saying, we know that what looks like a limit really is a limit, even though we can t prove it, or explain it, still we know it. RH: We know it the same way you know nobody has left a bear trap by your bedside. You just know it.

Mathematical Intuition: Poincaré, Pólya, Dewey 13 SS: Right. RH: But it s still possible that you re wrong. It is possible that something ridiculous actually happens. Not likely, not worth worrying about. But not impossible. SS: Then math is really just like everything else. What a bummer! I like math because it s not like everything else. In math, we know for sure. We prove things. One and one is two. Pi is irrational. A circle is round, not square. For sure. RH: Then why are you upset? Everything is just fine, isn t it? SS: Why don t you admit it? If you don t have a proof, you just don t know if L is the limit or not. RH: That s a fair question. So what is the answer? SS: Because you really want to think you know L is the limit, even if it s not true. RH: Not that it s not true, just that it might not be true. Thanks for your kind attention. What is supposed to be the meaning of this performance? What am I getting at? In this talk, I am NOT attempting to make a contribution to the problem of induction. Therefore, I may be allowed to omit a review of its 2,500-year literature. I am reporting and discussing what people really do, in practical convergence calculations, and in the process of mathematical discovery. I am going into a discussion of practical knowledge in mathematics, as a kind of real knowledge, even though it is not demonstrative or deductive knowledge. I try to explain why people must do what they do, in order to accomplish what they are trying to accomplish. I will conclude by arguing that the right broader context for the philosophy of mathematical practice is actually the philosophy of pragmatism, as expounded by John Dewey. But first of all, just this remarkable fact. What we do when we want actual numbers may be totally unjustified, according to our theory and our definition. And even more remarkable-nobody seems to notice, or to worry about it! Why is that? Well, the definition of convergence taught in calculus classes, as developed by those great men Augustin Cauchy and Karl Weierstrass, seems to actually convey what we want to mean by limit and convergence. It is a great success. Just look at the glorious edifice of mathematical analysis! On the other hand, in specific cases, it often is beyond our powers to give a rigorous error estimate, even when we have an approximation scheme that seems perfectly sound. As in the major problems of three-dimensional continuum mechanics with realistic nonlinearities, such as oceanography, weather prediction, stability of large complex structures like big bridges and airplanes... And even when we could possibly give a rigorous error estimate, it would often require great expenditure of time and labor. Surely it s OK to just use the result of a calculation when it makes itself evident and there s no particular reason to expect any hidden difficulty. In brief, we are virtually compelled by the practicalities to accept the number that computation seems to give us, even though, by the standards of rigorous logic, there is still an admitted possibility that we may be mistaken. This computational result is a kind of mathematical knowledge! It is practical knowledge, knowledge sound

14 R. Hersh enough to be the basis of practical decisions about things like designing bridges and airplanes-matters of life and death. In short, I am proclaiming that in mathematics, apart from and distinct from the so-called deductive or demonstrative knowledge, there is also ordinary, fallible knowledge, of the same sort as our daily knowledge of our physical environment and our own bodies. Anything new that we learn about the world involves plausible reasoning, which is the only kind of reasoning for which we care in everyday affairs, cf. [19]. This sentence of Pólya s makes an implicit separation between mathematics and everyday affairs. But nowadays, in many different ways, for many different kinds of people, mathematics blends into every-day affairs. In these situations, the dominance of plausible over demonstrative reasoning applies even to mathematics itself, as in the daily labors of numerical analysts, applied mathematicians, design engineers...controlling a rocket trip to the moon is not an exercise in mathematical rigor. It relies on a lack of malice on the part of that Being referred to by Albert Einstein as der lieber Gott. (For fear of misunderstanding, I explain this is not a confession of belief in a Supreme Being. It s just Einstein s poetic or metaphoric way of saying, Nature is not an opponent consciously trying to trick us.) But it s not only that we have no choice in the matter. It s also that, to tell the truth, it seems perfectly reasonable! Believing what the computation tells us is just what people have been doing all along, and (nearly always) it does seem to be OK. What s wrong with that? This kind of reasoning is sometimes called plausible, and sometimes called intuitive. I will say a little more about those two words pretty soon. But I want to draw your attention very clearly to two glaring facts about this kind of plausible or intuitive reasoning. First of all, it is pretty much the kind of reasoning that we are accustomed to in ordinary empirical science, and in technology, and in fact in everyday thinking, dealing with any kind of practical or realistic problem of human life. Secondly, it makes no claim to be demonstrative, or deductive, or conclusive, as is often said to be the essential characteristic of mathematical thinking. We are face to face with mathematical knowledge that is not different in kind from ordinary everyday commonplace human knowledge. Fallible! But knowledge, nonetheless! Never mind the pretend doubt of philosophical skepticism. We are adults, not infants. Human adults know a lot! How to find their way from bed to breakfast-and people s names and faces and so forth and so on. This is real knowledge. It is not infallible, not eternal, not heavenly, not Platonic, it is just what daily life depends on, that s all. That s what I mean by ordinary, practical, everyday knowledge. Based not mainly on rigorous demonstration or deduction, but mainly on experience properly interpreted. And here we see mathematical knowledge that is of the same ordinary, everyday kind, based not on infallible deduction, but on fallible, plausible, intuitive thinking. Then what justifies it in a logical sense? That is, what fundamental presupposition about the world, about reality, lies behind our willingness to commit this logical offense, of believing what isn t proved?

Mathematical Intuition: Poincaré, Pólya, Dewey 15 I have already quoted the famous saying of Albert Einstein that supplies the key to unlocking this paradox. My friend Peter Lax supplied the original German, I only remembered the English translation. Raffinniert ist der lieber Gott, aber boshaft ist Er nicht. The Lord God is subtle, but He is not malicious. Of course, Einstein was speaking as a physicist struggling to unravel the secrets of Nature. The laws of Nature are not always obvious or simple, they are often subtle. But we can believe, we must believe, that Nature is not set up to trick us, by a malicious opponent. God, or Nature, must be playing fair. How do we know that? We really don t know it, as a matter of certainty! But we must believe it, if we seek to understand Nature with any hope of success. And since we do have some success in that search, our belief that Nature is subtle but not malicious is justified. This problem of inferring generalizations from specific instances is known in logic as the problem of induction. My purpose is to point out that such generalizations in fact are made, and must be made, not only in daily life and in empirical science, but also in mathematics. That is, in the practice of mathematics also we must believe that we are not dealing with a malicious opponent who is seeking to trick us. We experiment, we calculate, we draw diagrams. And eventually, using caution and the experience of the ages, we see the light. Gauss famously said, I have my theorems. Now I have to find my proofs. But is it not naïve, for people who have lived through the hideous twentieth century, to still hope that God is not malicious? Consider, for example, people who for thousands of years have lived safely on some atoll in the South Pacific. Today an unforeseen tsunami drowns them all. Might they not curse God in their last breath? Here is an extensive quote from Leonhard Euler, by way of George Pólya. Euler is speaking of a certain beautiful and surprising regularity in the sum of the divisors of the integers. This law, which I shall explain in a moment, is in my opinion, so much more remarkable as it is of such a nature that we can be assured of its truth without giving it a perfect demonstration. Nevertheless, I shall present such evidence for it as might be regarded as almost equivalent to a rigorous demonstration... Anybody can satisfy himself of its truth by as many examples as he may wish to develop. And since I must admit that I am not in a position to give it a rigorous demonstration, I will justify it by a sufficiently large number of examples... I think these examples are sufficient to discourage anyone from imagining that it is by mere chance that my rule is in agreement with the truth... The examples that I have just developed will undoubtedly dispel any qualms which we might have had about the truth of my formula... It seems impossible that the law which has been discovered to hold for 20 terms, for example, would not be observed in the terms that follow. (Taken from [19].)

16 R. Hersh Observe two things about this quote from Euler. First of all, for him the plausible reasoning in this example is so irresistible that it leaves no room for doubt. He is certain that anyone who looks at his examples is bound to agree. Yet secondly, he strongly regrets his inability to provide a demonstration of the fact, and still hopes to find one. But since he is already certain of the truth of his finding, why ask for a demonstrative proof? The answer is easy, for anyone familiar with mathematical work. The demonstration would not just affirm the truth of the formula, it would show why the formula MUST be true. That is the main importance of proof in mathematics! A plausible argument, relying on examples, analogy and induction, can be very strong, can carry total conviction. But if it is not demonstrative, it fails to show why the result MUST be true. That is to say, it fails to show that it is rigidly connected to established mathematics. At the head of Chap. V of [19], Pólya placed the following apocryphal quotation, attributed to the traditional mathematics professor : When you have satisfied yourself that the theorem is true, you start proving it. This faith that experience is not a trap laid to mislead us is the unstated axiom. It lets us believe the numbers that come out of our calculations, including the canned programs that engineers use every day as black boxes. We know that it can sometimes be false. But even as we keep possible tsunamis in mind, we have no alternative but to act as if the world makes sense. We must continue to act on the basis of our experience. (Including, of course, experiences of unexpected disasters.) Consider this recollection of infantile mathematical research by the famous physicist Freeman Dyson, who wrote in 2004: One episode I remember vividly, I don t know how old I was; I only know that I was young enough to be put down for an afternoon nap in my crib... I didn t feel like sleeping, so I spent the time calculating. I added one plus a half plus a quarter plus an eighth plus a sixteenth and so on, and I discovered that if you go on adding like this forever you end up with two. Then I tried adding one plus a third plus a ninth and so on, and discovered that if you go on adding like this forever you end up with one and a half. Then I tried one plus a quarter and so on, and ended up with one and a third. So I had discovered infinite series. I don t remember talking about this to anybody at the time. It was just a game I enjoyed playing. Yes, he knew the limit! How did he know it? Not the way we teach it in high school (by getting an exact formula for the sum of n terms of a geometric sequence, and then proving that as n goes to infinity, the difference from the proposed limit becomes and remains arbitrarily small.) No, just as when we first show this to tenthgraders, he saw that the sums follow a simple pattern that clearly is converging to 2. The formal, rigorous proof gives insight into the reason for a fact we have already seen plainly. Can we go wrong this way? Certainly we can. Another quote from Euler.

Mathematical Intuition: Poincaré, Pólya, Dewey 17 There are even many properties of the numbers with which we are well acquainted, but which we are not yet able to prove; only observations have led us to their knowledge..., the kind of knowledge which is supported only by observations and is not yet proved must be carefully distinguished from the truth; it is gained by induction, as we usually say. Yet we have seen cases in which mere induction led to error. Therefore, we should take great care not to accept as true such properties of the numbers which we have discovered by observation and which are supported by induction alone. Indeed, we should use such a discovery as an opportunity to investigate more exactly the properties discovered and to prove or disprove them. (In [19], p. 3.) Notice how Euler distinguishes between knowledge and truth! He does say knowledge, not mere conjecture. There is a famous theorem of Littlewood concerning a pair of number-theoretic functions π(x) and Li(x). All calculation shows that Li(x) is greater than π(x),for x as large as we can calculate. Yet Littlewood proved that eventually π(x) becomes greater than Li(x), and not just once, but infinitely often! Yes, mathematical truth can be very subtle. While trusting it not to be malicious, we must not underestimate its subtlety. (π(x) is the prime counting function and Li(x) is the logarithmic integral function.) 3 Mathematical Intuition We are concerned with the philosophy of mathematical practice. Mathematical practice includes studying, teaching and applying mathematics. But I suppose we have in mind first of all the discovery and creation of mathematics-mathematical research. We start with Jacques Hadamard, go on to Henri Poincaré, move on to George Pólya, and then to John Dewey. Hadamard had a very long life and a very productive career. His most noted achievement (shared independently by de la Vallée Poussin) was proving the logarithmic distribution of the prime numbers. I want to recall a famous remark of Hadamard s. The object of mathematical rigor is to sanction and legitimize the conquests of intuition, and there never was any other object for it. (See [20].) From the viewpoint of standard philosophy of mathematics, this is a very surprising, strange remark. Isn t mathematical rigor-that is, strict deductive reasoningthe most essential feature of mathematics? And indeed, what can Hadamard even mean by this word, intuition? A word that means one thing to Descartes, another thing to Kant. I think the philosophers of mathematics have pretty unanimously chosen to ignore this remark of Hadamard. Yet Hadamard did know a lot of mathematics, both rigorous and intuitive. And this remark was quoted approvingly by both Borel and Pólya. It seems to me that this bewildering remark deserves to be taken seriously. Let s pursue the question a step further, by recalling the famous essay Mathematical Discovery, written by Hadamard s teacher, Henri Poincaré, cf. [16]. Poincaré

18 R. Hersh was one of the supreme mathematicians of the turn of the nineteenth and twentieth century. We ve been hearing his name recently, in connection with his conjecture on the 3-sphere, just recently proved by Grisha Perelman of St. Petersburg. Poincaré was not only a great mathematician, he was a brilliant essayist. And in the essay Mathematical Discovery, Poincaré makes a serious effort to explain mathematical intuition. He tells the famous story of how he discovered the Fuchsian and Theta- Fuchsian functions. He had been struggling with the problem unsuccessfully when he was distracted by being called up for military service: At this moment I left Caën, where I was then living, to take part in a geological conference arranged by the School of Mines. The incidents of the journey made me forget my mathematical work. When we arrived at Coutances, we got into a bus to go for a drive, and, just as I put my foot on the step the idea came to me, though nothing in my former thoughts seemed to have prepared me for it, that the transformations I had used to define Fuchsian functions were identical with those of non-euclidean geometry. I made no verification, and had no time to do so, since I took up the conversation again as soon as I had sat down in the bus, but I felt absolute certainty at once. When I got back to Caën, I verified the result at my leisure to satisfy my conscience. (Cf. [16].) What a perfect example of rigor merely legitimizing the conquests of intuition! How does Poincaré explain it? First of all, he points out that some sort of subconscious thinking must be going on. But if it is subconscious, he presumes it must be running on somehow at random. How unlikely, then, for it to find one of the very few good combinations, among the huge number of useless ones! To explain further, he writes: If I may be permitted a crude comparison, let us represent the future elements of our combinations as something resembling Epicurus hooked atoms. When the mind is in complete repose these atoms are immovable; they are, so to speak, attached to the wall... On the other hand, during a period of apparent repose, but of unconscious work, some of them are detached from the wall and set in motion. They plough through space in all directions, like a swarm of gnats, for instance, or, if we prefer a more learned comparison, like the gaseous molecules in the kinetic theory of gases. Their mutual impacts may then produce new combinations. (Cf. [16].) The preliminary conscious work detached them from the wall. The mobilized atoms, he speculated, would therefore be those from which we might reasonably expect the desired solution... My comparison is very crude, but I cannot well see how I could explain my thought in any other way. (Cf. [16].) What can we make of this picture of Epicurean hooked atoms, flying about somewhere-in the mind? A striking, suggestive image, but one not subject even in principle to either verification or disproof. Our traditional philosopher remains little interested. This is fantasy or poetry, not science or philosophy. But this is Poincaré! He knows what he s talking about. He has something important to tell us. It s not easy to understand, but let s take him seriously, too.

Mathematical Intuition: Poincaré, Pólya, Dewey 19 To be fair, Poincaré proposed his image of gnats or gas molecules only after mentioning the possibility that the subconscious is actually more intelligent than the conscious mind. But this, he said, he was not willing to contemplate. However, other writers have proposed that the subconscious is less inhibited, more imaginative, more creative than the conscious. (Poincaré s essay title is sometimes translated as Mathematical Creation rather than Mathematical Discovery.) David Hilbert supposedly once said of a student who had given up mathematics for poetry, Good! He didn t have enough imagination for mathematics. Hadamard [12] in 1949 carefully analyzed the role of the subconscious in mathematical discovery and its connection with intuition. It is time for contemporary cognitive psychology to pay attention to Hadamard s insights. See the reference quoted in the Appendix about current scientific work on the creative power of the subconscious. Before going on, I want to mention the work of Carlo Cellucci, Emily Grosholz, and Andrei Rodin. Cellucci strongly favors plausible reasoning, but he rejects intuition. However, the intuition he rejects isn t what I m talking about. He s rejecting the old myth, of an infallible insight straight into the Transcendental. Of course, I m not advocating that outdated myth. Emily Grosholz, on the other hand, takes intuition very seriously. Her impressive historical study of what she calls internal intuition is in the same direction as my own thinking being presented here. Andrei Rodin has recently written a remarkable historical study of intuition, see [24]. He shows that intuition played a central role in Lobachevski s non-euclidean geometry, in Zermelo s axiomatic set theory, and even in up-to-date category theory. (By the way, in category theory he could also have cited the standard practice of proof by diagram chasing as a blatant example of intuitive, visual proof.) His exposition makes the indispensable role of intuition clear and convincing. But his use of the term intuition remains, one might say, intuitive, for he offers no definition of the term, nor even a general description, beyond his specific examples. 4 Pólya My most helpful authority is George Pólya. I actually induced Pólya to come give talks in New Mexico, for previously, as a young instructor, I had met him at Stanford where he was an honored and famous professor. Pólya was not of the stature of Poincaré or Hilbert, but he was still one of the most original, creative, versatile, and influential mathematicians of his generation. His book with Gábor Szegö [21] made them both famous. It expounds large areas of advanced analytic function theory by means of a carefully arranged, graded sequence of problems with hints and solutions. Not only does it teach advanced function theory, it also teaches problem-solving. And by example, it shows how to teach mathematics by teaching problem-solving. Moreover, it implies a certain view of the nature of mathematics, so it is a philosophical work in disguise. Later, when Pólya wrote his very well-known, influential books on mathematical heuristic, he admitted that what he was doing could be regarded as having philo-

20 R. Hersh sophical content. He writes, I do not know whether the contents of these four chapters deserve to be called philosophy. If this is philosophy, it is certainly a pretty low-brow kind of philosophy, more concerned with understanding concrete examples and the concrete behavior of people than with expounding generalities. (Cf. [19], page viii.) Unpretentious as Pólya was, he was still aware of his true stature in mathematics. I suspect he was also aware of the philosophical depth of his heuristic. He played it down because, like most mathematicians (I can only think of one or two exceptions), he disliked controversy and arguing, or competing for the goal of becoming top dog in some cubbyhole of academia. The Prince of Mathematicians, Carl Friedrich Gauss, kept his monumental discovery of non-euclidean geometry hidden in a desk drawer to avoid stirring up the Bœotians, as he called them, meaning the post-kantian German philosophy professors of his day. (In ancient Athens, Bœotian was slang for ignorant country hick.) Raymond Wilder was a leading topologist who wrote extensively on mathematics as a culture. He admitted to me that his writings implicitly challenged both formalism and Platonism. Why not say so? I asked. Because he didn t relish getting involved in philosophical argument. Well, how does Pólya s work on heuristic clarify mathematical intuition? Pólya s heuristic is presented as pedagogy. Pólya is showing the novice how to solve problems. But what is solving a problem? In the very first sentence of the preface to [20], he writes: Solving a problem means finding a way out of a difficulty, a way around an obstacle, attaining an aim which was not immediately attainable. Solving problems is the specific achievement of intelligence, and intelligence is the specific gift of mankind: solving problems can be regarded as the most characteristically human activity. Problem is simply another word for any project or enterprise which one cannot immediately take care of with the tools at hand. In mathematics, something more than a mere calculation. Showing how to solve problems amounts to showing how to do research! Pólya s exposition is never general and abstract, he always uses a specific mathematical problem for the heuristic he wants to teach. His mathematical examples are always fresh and attractive. And his heuristic methods? First of all, there is what he calls induction. That is, looking at examples, as many as necessary, and using them to guess a pattern, a generalization. But be careful! Never just believe your guess! He insists that you must Guess and test, guess and test. Along with induction, there is analogy, and there is making diagrams, graphs and every other kind of picture, and then reasoning or guessing from the picture. And finally, there s the default hypothesis of chance that an observed pattern is mere coincidence. (Mark Steiner has the distinction among philosophers of paying serious attention to Pólya. After quoting at length from Pólya s presentation of Euler s heuristic derivation of the sum of a certain infinite series, Steiner comes to an important conclusion: In mathematics we can have knowledge without proof! Based on the testimony of mathematicians, he even urges philosophers to pay attention to the question of mathematical intuition.) I have two comments about Pólya s heuristic that I think he would have accepted. First of all, the methods he is presenting, by means of elementary examples, are methods he used himself in research. In fact, my main source was my own research, and my treatment of many an elementary problem mirrors my experience

Mathematical Intuition: Poincaré, Pólya, Dewey 21 with advanced problems. (Cf. [20], page xi.) In teaching us how to solve problems, he s teaching us about mathematical practice: How it works. What is done. To find out What is mathematics? we must simply reinterpret Pólya s examples as descriptive rather than pedagogical. Secondly, with hardly any stretching or adjustment, the heuristic devices that he s teaching can be applied for any other kind of problem-solving, far beyond mathematics. He actually says that he is bringing to mathematics the kind of thinking ordinarily associated with empirical science. But we can go further. These ways of thinking are associated with every kind of problem-solving, in every area of human life! Someone needed to get across a river or lake and had the brilliant idea of a boat whether it was a dugout log or a birch bark canoe. Someone else, needing shelter from the burning sun in the California Mojave, thought of digging a hole in the ground. And someone else, under the piercing wind of northern Canada, thought of making a shelter from blocks of ice. How does anyone think of such a thing, solve such a concrete problem? By some kind of analogy with something else he has seen, or perhaps been told about. By plausible thinking. And often by a sudden insight that arises from below. Intuitively, you might say. 5 Mental Models It often happens that a concrete problem, whether in science or in ordinary daily life, is pressing on the mind, even when the particular materials or objects in question are not physically present. You keep on thinking about it, while you re walking, and when you re waking from sleep. Productive thought commonly takes place, in the absence of the concrete objects or materials being thought about. This thinking about something not present to sight or touch can be called abstract thinking. Abstract thinking about a concrete object. How does that work? How can our mind/brain think productively about something that s not there in front of the eyes? Evidently, it operates on something mental, what we may call a mental image or representation. In the current literature of cognitive psychology, one talks about a mental model. In this article, I use the term mental model to mean a mental structure built from recollected facts (some expressed in words), along with an ensemble of sensory memories, perhaps connected, as if by walking around the object in question, or by imagining the object from underneath or above, even if never actually seen in these views. A rich complex of connected knowledge and conjecture based on verbal, visual, kinesthetic, even auditory or olfactory information, but simplified, to exclude irrelevant details. Everything that s helpful for thinking about the object of interest when the object isn t here. Under the pressure of a strong desire or need to solve a specific problem, we assemble a mental model which the mind-brain can manipulate or analyze. Subconscious thinking is not a special peculiarity of mathematical thinking, but a common, taken-for-granted, part of every-day problem-solving. When we consider

22 R. Hersh this commonplace fact, we aren t tempted to compare it to a swarm of gnats hooking together at random. No, we assume, as a matter of course, that this subconscious thinking follows rules, methods, habits or pathways, that somehow, to some extent, correspond to the familiar plausible thinking we do when we re wide awake. Such as thinking by analogy or by induction. After all, if it is to be productive, what else can it do? If it had any better methods, then those better methods would also be what we would follow in conscious thinking! And subconscious thinking in mathematics must be much like subconscious thinking in any other domain, carrying on plausible reasoning as enunciated by various writers, above all by George Pólya. This description of subconscious thinking is not far from Michael Polanyi s tacit dimension. When applied to everyday problem solving, all this is rather obvious, perhaps even banal. My goal is to clarify mathematical intuition, in the sense of Hadamard and Poincaré. Intuition in the sense of Hadamard and Poincaré is a fallible psychological experience that has to be accounted for in any realistic philosophy of mathematics. It simply means guesses or insights attained by plausible reasoning, either fully conscious or partly subconscious. In this sense, it is a specific phenomenon of common experience. It has nothing to do with the ancient mystical myth of an intuition that surpasses logic by making a direct connection to the Transcendental. The term abstract thinking is commonplace in talk about mathematics. The triangle, the main subject of Euclidean geometry, is an abstraction, even though it s idealized from visible triangles on the blackboard. Thinking of a physical object in its absence, like a stream to be crossed or a boat to be imagined and then built, is already abstract thinking, and the word abstract connects us to the abstract objects of mathematics. Let me be as clear and simple as I can be about the connection. After we have some practice drawing triangles, we can think about triangles, we discover properties of triangles. We do this by reasoning about mental images, as well as images on paper. This is already abstract thinking. When we go on to regular polygons of arbitrarily many sides, we have made another departure. Eventually, we think of the triangle as a 2-simplex, and abstract from the triangle to the n-simplex. For n = 3 this is just the tetrahedron, but for n = 4, 5, or 6, it is something never yet seen by human eye. Yet these higher simplexes also can become familiar and, as it were, concrete-seeming. If we devote our waking lives to thinking about them, then we have some kind of mental model of them. Having this mental model, we can access it, and thereby we can reason intuitively-have intuitive insights-by which I mean simply insights not based on consciously known reasoning. An intuition is then simply a belief (possibly mistaken!) arising from internal inspection of a mental image or representation a model. It may be assisted by subconscious plausible reasoning, based on the availability of that mental image. We do this in practical life. We do it in empirical science, and in mathematics. In empirical science and ordinary life, the image may stand for either an actual object, a physical entity, or a potential one that could be realized physically. In mathematics, our mental model is sometimes idealized from a physical object-for example, from a collection of identical coins or buttons when we re thinking about arithmetic. But in mathematics we

Mathematical Intuition: Poincaré, Pólya, Dewey 23 also may possess a mental model with no physical counterpart. For example, it is generally believed that Bill Thurston s famous conjectures on the classification of four-manifolds were achieved by an exceptional ability, on the part of Thurston, to think intuitively in the fourth dimension. Perhaps Grisha Perelman was also guided by some four-dimensional intuition, in his arduous arguments and calculations to prove the Thurston program. To summarize, mathematical intuition is an application of conscious or subconscious heuristic thinking of the same kind that is used every day in ordinary life by ordinary people, as well as in empirical science by scientists. This has been said before, by both Hadamard and Pólya. In fact, this position is similar to Kurt Gödel s, who famously wrote, I don t see any reason why we should have less confidence in this kind of perception, i.e., in mathematical intuition, than in sense perception. Why, indeed? After all, both are fallible, but both are plausible, and must be based on plausible reasoning. For Gödel, however, as for every writer in the dominant philosophy of mathematics, intuition is called in only to justify the axioms. Once the axioms are written down, the role of mathematical intuition is strictly limited to heuristic to formulating conjectures. These await legitimation by deductive proof, for only deductive proof can establish certainty. Indeed, this was stated as firmly by Pólya as by any analytic philosopher. But what is meant by mathematical certainty? If it simply means deductive proof, this statement is a mere circular truism. However, as I meant to suggest by the little dialog at the beginning of this paper, there is also practical certainty, even within mathematics! We are certain of many things in ordinary daily life, without deductive proof, and this is also the case in mathematics itself. Practical certainty is a belief strong enough to lead to serious practical decisions and actions. For example, we stake our lives on the numerical values that went into the engineering design of an Airbus or the Golden Gate Bridge. Mainstream philosophy of mathematics does not recognize such practical certainty. Nevertheless, it is an undeniable fact of life. It is a fact of life not only in applied mathematics but also in pure mathematics. For example, the familiar picture of the Mandelbrot set, a very famous bit of recent pure mathematics, is generated by a machine computation. By definition, any particular point in the complex plane is inside the Mandelbrot set if a certain associated iteration stays bounded. If that iteration at some stage produces a number with absolute value greater than two, then, from a known theorem, we can conclude that the iteration goes to infinity, and the parameter point in question is outside the Mandelbrot set. What if the point is inside the Mandelbrot set? No finite number of iterations in itself can guarantee that the iteration will never go beyond absolute value 2. If we do eventually decide that it looks like it will stay bounded, we may be right, but we are still cheating. This decision is opportunistic and unavoidable, just as in an ordinary calculation about turbulent flow. Computation (numerics) is accepted by purists only as a source of conjectures awaiting rigorous proof. However, from the pragmatic, non-purist viewpoint, if numerics is our guide to action, then it is in effect a source of knowledge. Dewey called it warranted assertibility. (Possibly even a truth. A truth that remains open to possible reconsideration.)

24 R. Hersh Fig. 1 400 million polynomial roots Another example from pure mathematics appeared on John Baez s blog ([1]) where it is credited to Sam Derbyshire. His pictures plot the location in the complex plane of the roots of all polynomials of degree 24 with coefficients plus one or minus one. The qualitative features of these pictures are absolutely convincing i.e., impossible to disbelieve. Baez wrote, That s 2 24 polynomials, and about 24 2 24 roots or about 400 million roots! It took Mathematica 4 days to generate the coordinates of the roots, producing about 5 gigabytes of data. (Figure 1 shows the part of the plot in the first quadrant, for complex roots with non-negative real and imaginary parts.) There is more information in this picture than can even be formulated as conjectures, let alone seriously attacked with rigor. Since indeed we cannot help believing them (perhaps only believing with 99.999 % credence) then (pragmatically) we give them warranted assertibility, just like my belief that I can walk out my door without encountering sudden death in one form or another. The distinction between rigorous math and plausible math, pure math and applied math, etc., becomes blurred. It is still visible, certainly, but not so sharp. It s a little fuzzy. Purely computational results in pure mathematics, when backed up by sophisticated checking against a relevant theory, have a factual status similar to that of accepted facts from empirical science. The distinction between what is taken to be known, and what is set aside as merely guessed or conjectured, is not so cut and dried as the usual discussions claim to believe.

Mathematical Intuition: Poincaré, Pólya, Dewey 25 6 Mental Models Subject to Social Control Plausible or heuristic thinking is applied, either consciously or subconsciously, to mental models. These mental models may correspond to tangible or visible physical objects in ordinary life and empirical science. Or they may not correspond to any such things, but may be pure mental representations, as in much of contemporary analysis, algebra, and even geometry. By pure mental models I mean models not obtained directly by idealization of visual or other sensual experience. But what controls these mental models? If they have no physical counterpart, what keeps them from being wildly idiosyncratic and incommunicable? What we have omitted up to this point, and what is the crux of the matter: mathematical images are not private, individual entities. From the origin of mathematics in bartering, buying and selling, or in building the Parthenon and the Pyramids, this subject has always been a social, an inter-subjective activity. Its advances and conquests have always been validated, corrected and absorbed in a social context-first of all, in the classroom. Mathematicians can and must talk to each other about their ideas. One way or the other, they do communicate, share and compare their conceptions of mathematical entities, which means precisely these models, these images and representations I have been describing. Discrepancies are recognized and worked out, either by correcting errors, reconciling differences, or splitting apart into different, independent pathways. Appropriate terminology and symbols are created as needed. Mathematics depends on a mutually acknowledging group of competent practitioners, whose consensus decides at any time what is regarded as correct or incorrect, complete or incomplete. That is how it always worked, and that is how it works today. This was made very clear by the elaborate process in which Perelman s proposed proof of the Thurston program (including the Poincaré conjecture) was vetted, examined, discussed, criticized, and finally accepted by the Ricci flow community, and then by its friends in the wider communities of differential geometry and low-dimensional topology, and then by the prize committees of the Fields Medal and the Clay Foundation. Thus, when we speak of a mathematical concept, we speak not of a single isolated mental image, but rather of a family of mutually correcting mental images. They are privately owned, but publicly checked, examined, corrected, and accepted or rejected. This is the role of the mathematical research community, how it indoctrinates and certifies new members, how it reviews, accepts or rejects proposed publication, how it chooses directions of research to follow and develop, or to ignore and allow to die. All these social activities are based on a necessary condition: that the individual members have mental models that fit together, that yield the same answers to test questions. A new branch of mathematics is established when consensus is reached about the possible test questions and their answers. That collection of possible questions and answers (not necessarily explicit) becomes the means of accepting or rejecting proposed new members. If two or three mathematicians do more than merely communicate about some mathematical topic, but actually collaborate to dig up new information and understanding about it, then the matching of their mental models must be even closer.